Genomic epidemiology of the Escherichia coli O104:H4
outbreaks in Europe, 2011
The MIT Faculty has made this article openly available. Please share
how this access benefits you. Your story matters.
Citation
Grad, Y. H. et al. “Genomic Epidemiology of the Escherichia Coli
O104:H4 Outbreaks in Europe, 2011.” Proceedings of the
National Academy of Sciences 109.8 (2012): 3065–3070.
Copyright ©2012 by the National Academy of Sciences
As Published
http://dx.doi.org/10.1073/pnas.1121491109
Publisher
National Academy of Sciences
Version
Final published version
Accessed
Thu May 26 08:49:44 EDT 2016
Citable Link
http://hdl.handle.net/1721.1/72563
Terms of Use
Article is made available in accordance with the publisher's policy
and may be subject to US copyright law. Please refer to the
publisher's site for terms of use.
Detailed Terms
Genomic epidemiology of the Escherichia coli O104:H4
outbreaks in Europe, 2011
Yonatan H. Grada,b, Marc Lipsitchb,c, Michael Feldgardend, Harindra M. Arachchid, Gustavo C. Cerqueirad,
Michael FitzGeraldd, Paul Godfreyd, Brian J. Haasd, Cheryl I. Murphyd, Carsten Russd, Sean Sykesd, Bruce J. Walkerd,
Jennifer R. Wortmand, Sarah Youngd, Qiandong Zengd, Amr Abouelleild, James Bochicchiod, Sara Chauvind,
Timothy DeSmetd, Sharvari Gujjad, Caryn McCowand, Anna Montmayeurd, Scott Steelmand, Jakob Frimodt-Møllere,f,
Andreas M. Petersenf,g, Carsten Struvef, Karen A. Krogfeltf, Edouard Bingenh,i, François-Xavier Weillj,
Eric S. Landerd,k,l,1, Chad Nusbaumd, Bruce W. Birrend, Deborah T. Hunga,d,m,n,2, and William P. Hanageb,1,2
a
Department of Medicine, Brigham and Women’s Hospital, Boston, MA 02115; bDepartment of Epidemiology, Center for Communicable Disease
Dynamics, and cDepartment of Immunology and Infectious Diseases, Harvard School of Public Health, Boston, MA 02115; dBroad Institute of Harvard and
Massachusetts Institute of Technology, Cambridge, MA 02142; eDepartment of Clinical Microbiology, Hillerød Sygehus, 3400 Hillerød, Denmark; fDepartment
of Microbial Surveillance and Research, Statens Serum Institute, 2300 Copenhagen, Denmark; gDepartment of Gastroenterology, Hvidovre University Hospital,
2650 Hvidovre, Denmark; hLaboratoire Associé au Centre National de Référence des Escherichia coli et Shigella, Service de Microbiologie, Hôpital Robert
Debré, Assistance Publique-Hôpitaux de Paris, 75019 Paris, France; iUniversité Paris-Diderot, Sorbonne Paris Cité, EA3105, 75505 Paris, France; jInstitut Pasteur,
Unité des Bactéries Pathogènes Entériques, Centre National de Référence des Escherichia coli et Shigella, 75015 Paris, France; kDepartment of Biology,
Massachusetts Institute of Technology, Cambridge, MA 02139; lDepartment of Systems Biology, and mDepartment of Microbiology and Immunobiology,
Harvard Medical School, Boston, MA 02115; and nDepartment of Molecular Biology, Massachusetts General Hospital, Boston, MA 02115
The degree to which molecular epidemiology reveals information
about the sources and transmission patterns of an outbreak
depends on the resolution of the technology used and the samples
studied. Isolates of Escherichia coli O104:H4 from the outbreak centered in Germany in May–July 2011, and the much smaller outbreak
in southwest France in June 2011, were indistinguishable by standard tests. We report a molecular epidemiological analysis using
multiplatform whole-genome sequencing and analysis of multiple
isolates from the German and French outbreaks. Isolates from the
German outbreak showed remarkably little diversity, with only
two single nucleotide polymorphisms (SNPs) found in isolates from
four individuals. Surprisingly, we found much greater diversity (19
SNPs) in isolates from seven individuals infected in the French outbreak. The German isolates form a clade within the more diverse
French outbreak strains. Moreover, five isolates derived from a single infected individual from the French outbreak had extremely
limited diversity. The striking difference in diversity between the
German and French outbreak samples is consistent with several
hypotheses, including a bottleneck that purged diversity in the
German isolates, variation in mutation rates in the two E. coli outbreak populations, or uneven distribution of diversity in the seed
populations that led to each outbreak.
|
food-borne outbreak Shiga toxin
enterohemorrhagic E. coli
| enteroaggregative E. coli |
I
n May–July 2011, two outbreaks of bloody diarrhea and hemolytic uremic syndrome (HUS) occurred in Europe: one
centered in Germany (around 4,000 cases of bloody diarrhea, 850
cases of HUS and 50 deaths), and a much smaller outbreak in
southwest France, near Bordeaux (15 cases of bloody diarrhea, 9
of which progressed to HUS) (1–4). Both outbreaks were caused
by a strain of Shiga toxin-producing Escherichia coli of serotype
O104:H4 (2, 5), which possesses a plasmid, pAA, characteristic
of enteroaggregative E. coli, as well as a plasmid encoding an
extended-spectrum β-lactamase (ESBL) (3). The proportion of
patients infected with E. coli O104:H4 who develop complications, including HUS, is higher than seen in prior outbreaks (1, 6).
The source of the outbreaks was epidemiologically linked to
contaminated sprouts, and evidence indicates the outbreaks are
connected to a 15,000-kg seed shipment from Egypt that arrived in
Germany in December 2009. The majority of the seeds from the
shipment (10,500 kg) was then sent to a German seed distributor,
which supplied the implicated German sprout farm. Four hundred
kilograms of the original seed shipment was sent to an English
www.pnas.org/cgi/doi/10.1073/pnas.1121491109
seed distributor, which then repacked seeds into 50-g packets
passed on to French garden stores (7). The seeds from a packet
were then germinated into sprouts at a children’s community
center, and the sprouts were served on June 8, 2011, leading to the
French outbreak (2).
Epidemiological investigations of outbreaks aim to combine
various approaches to reconstruct in detail the chain of events
that led to the outbreak. In principle, genetic information, such
as the patterns of genetic diversity among isolates, can aid in
tracking the origins and transmission of the pathogens. Genetic
diversity can indicate how long the pathogenic lineage has been
diversifying and shed light on when, where, and how this E. coli
originated and entered the human food chain. In practice, such
inferences require extensive and highly accurate genetic information. Even small error rates, which matter little for comparing
an outbreak strain to historical isolates, could obscure genuine
phylogenetic signal in comparing extremely closely related genomes
from within an outbreak.
Based on conventional molecular epidemiological characterization (including virulence gene content, serotyping, multilocus
sequence typing, rep-PCR, pulsed-field gel electrophoresis, optical
mapping, and antimicrobial susceptibility testing), the outbreak
strains in Germany and France appear identical (2, 8) (see also
Author contributions: Y.H.G., M.L., C.R., C.N., B.W.B., D.T.H., and W.P.H. designed research; Y.H.G., H.M.A., G.C.C., M. FitzGerald, P.G., B.J.H., C.R., S. Sykes, B.J.W., J.R.W.,
S.Y., Q.Z., A.A., J.B., S.C., T.D., S.G., C. McCowan, A.M., S. Steelman, E.B., F.-X.W., and C.N.
performed research; Y.H.G., C.R., S. Sykes, B.J.W., J.R.W., S.Y., Q.Z., J.F.-M., A.M.P., C.S., K.A.K.,
E.B., and F.-X.W. contributed new reagents/analytic tools; Y.H.G., M.L., M. Feldgarden,
M. FitzGerald, P.G., B.J.H., C.R., S. Sykes, B.J.W., J.R.W., S.Y., Q.Z., F.-X.W., E.S.L., C.N.,
B.W.B., D.T.H., and W.P.H. analyzed data; and Y.H.G., M.L., M. Feldgarden, C. Murphy,
C.R., J.R.W., K.A.K., E.B., F.-X.W., E.S.L., C.N., B.W.B., D.T.H., and W.P.H. wrote the paper.
The authors declare no conflict of interest.
Freely available online through the PNAS open access option.
Data deposition: The sequences reported in this paper have been deposited in the GenBank database (accession nos. E. coli C227-11 AFRH01000000; E. coli C236-11
AFRI01000000; E. coli Ec04-8351 AFRL01000000; E. coli Ec11-3677 AFRM01000000; E. coli
Ec09-7901 AFRK01000000; E. coli Ec11-4632-C1 AFVA01000000; E. coli Ec11-4632-C2
AFVB01000000; E. coli Ec11-4632-C3 AFVC01000000; E. coli Ec11-4632-C4 AFVD01000000;
E. coli Ec11-4632-C5 AFVE01000000; E. coli Ec11-4404 AFUX01000000; E. coli Ec11-4522
AFUY01000000; E. coli Ec11-4623 AFUZ01000000; E. coli Ec11-5538 AFVF01000000; E. coli
Ec11-5537 AFVG01000000; and E. coli Ec11-5536 AFVH01000000).
1
To whom correspondence may be addressed. E-mail: lander@broadinstitute.org or
whanage@hsph.harvard.edu.
2
D.T.H. and W.P.H. contributed equally to this work.
This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.
1073/pnas.1121491109/-/DCSupplemental.
PNAS | February 21, 2012 | vol. 109 | no. 8 | 3065–3070
MICROBIOLOGY
Contributed by Eric S. Lander, December 29, 2011 (sent for review December 2, 2011)
SI Materials and Methods). However, these approaches do not
assess the full diversity among strains. A comprehensive strategy
requires whole-genome sequencing with accurate resolution on
the single nucleotide level and can be augmented by analysis of
gene and plasmid content.
Results
We first performed whole-genome sequencing using the Illumina
sequencing platform on four isolates from the outbreak centered
in Germany (Table 1). Among these four isolates, we found only
two SNPs relative to a published genome from the German
outbreak, TY2482 (9): two of the isolates showed no differences
relative to the reference, and two showed one SNP each (nucleotide positions 224851 and 1096014) (Table 2; see also SI
Materials and Methods, Table S1, and Fig. S1). We independently
confirmed the two SNPs by Sanger sequencing. As further validation of the sequence quality, we performed genome sequencing, assembly, and SNP calling of two of these isolates (C236-11
and C227-11), using an independent genome-sequencing technology (454 sequencing platform); this analysis found the same
two SNPs and no additional ones (see SI Materials and Methods
and Tables, S2, S3, and S4). Our observation of limited diversity
in the German outbreak isolates is consistent with a recent report that found no SNPs in two independent isolates from the
German outbreak (10).
We then analyzed strains from the smaller French outbreak.
We performed whole-genome sequencing on 11 isolates from
seven patients, including five isolated simultaneously from a single patient (Table 1). Surprisingly, the diversity of the isolates
from the French outbreak was considerably greater than that
from the German outbreak (Table 2). We found 19 SNPs, all of
which were validated by Sanger sequencing.
The five isolates from the single host showed virtually no
variation. Four isolates were identical, but the fifth lacked one
SNP shared by the other four (Fig. 1A and Table 2). Technically,
the low diversity within a single individual further confirms the
sequencing quality. Scientifically, it suggests that infection may
have involved a small inoculum [similar to the estimated low
infectious dose of E. coli O157:H7 (11)], or that a small number
of genotypes dominate within a host during an infection.
A maximum-likelihood phylogeny of the outbreak isolates
(Fig. 1A), rooted on historical E. coli O104:H4 isolates from 2004
and 2009 that we had also sequenced, showed that the limited
diversity seen in the samples from the large German outbreak was
nested within the greater diversity of French isolates. One SNP, at
location 1568661, distinguishes the historical 2004 and 2009 isolates and all but two of the French isolates from the outbreak
isolates from Germany. The most parsimonious explanation is
that the isolates from the outbreak in Germany represent a subset
of diversity seen in the French outbreak. We additionally placed
the outbreak isolates into broader phylogenetic context using
C227-11 as representative of the outbreak: historical E. coli O104:
H4 isolates 55989 [isolated from an HIV-positive adult from the
Central African Republic in the 1990s that, like the other isolates,
is enteroaggregative, but, in contrast, is not Shiga toxin-producing
(12)], 01–09591 [isolated from an individual in Germany in 2001
(13)], and the 2004 and 2009 isolates from individuals in France
and a commensal E. coli genome E1167 (Fig. 1B). Although the
historical E. coli O104:H4 isolates from 2001, 2004, and 2009 are
related to this outbreak, they do not appear to be ancestral.
To confirm that the diversity found in the French outbreak was
absent in the German outbreak, we analyzed sequence data from
eight additional German outbreak strains recently deposited in
GenBank (GOS1, GOS2, H112180540, H112180541, H112180280,
H112180282, H112180283, and LB226692). Although these genome
sequences are not suitable for de novo SNP prediction using our
approach (most lack quality scores), they can be evaluated for the
presence of known SNPs. We found that none of these genomes
contained any of the 19 SNPs seen in the French outbreak or the
two identified in the German outbreak (see SI Materials and
Methods for details), indicating that they share the same sequence
as TY2482 at these sites.
The identity of the SNPs suggests that they reflect recent diversification without evidence for either purifying or positive
selection (14). Specifically, the SNPs are not biased toward
protein-altering substitutions. Of the 21 SNPS, 3 (14.3%) SNPs
are intergenic (in keeping with the range of 12.3–13.8% of the
genome predicted to be intergenic) (Table S5). Of the 14 SNPs
within coding regions, 4 (28.6%) are synonymous.
We found that all German and French outbreak isolates contained the three plasmids, including pAA, the ESBL plasmid, and
a much smaller third plasmid, all of which have been identified in
other descriptions of the O104:H4 outbreak isolates (9, 10, 13, 15).
Through synteny and ortholog analysis, we computationally
predicted only one region of gene difference, a deletion in Ec115538, one of the French outbreak isolates (see SI Materials and
Methods for details). We confirmed the absence of an 836-bp
region in this genome by PCR analysis and note that it is adjacent
to an insertion sequence. This deleted region includes three
predicted genes and the 5′ end of a fourth predicted gene (SI
Table 1. E. coli O104:H4 isolates sequenced and analyzed in this study
Isolate name
Date of symptoms
Date of isolation
Age
Sex
Clinical manifestations
Outbreak
Ec04-8351
Ec09-7901
Ec11-3677
Ec11-3798
C227-11
C236-11
Ec11-4404
Ec11-5536
Ec11-5537
Ec11-5538
Ec11-4632_C1
Ec11-4632_C2
Ec11-4632_C3
Ec11-4632_C4
Ec11-4632_C5
Ec11-4623
Ec11-4522
2004
2009
May 21, 2011
May 21, 2011
May 14, 2011
May 19, 2011
June 17, 2011
June 17, 2011
June 20, 2011
June 20, 2011
June 15, 2011
June 15, 2011
June 15, 2011
June 15, 2011
June 15, 2011
June 18, 2011
June 18, 2011
2004
2009
May 26, 2011
May 25, 2011
May 18, 2011
May 21, 2011
June 21, 2011
June 24, 2011
June 24, 2011
June 24, 2011
June 25, 2011
June 25, 2011
June 25, 2011
June 25, 2011
June 25, 2011
June 27, 2011
June 22, 2011
50
6
31
55
64
23
42
49
35
41
47
47
47
47
47
31
65
Male
Male
Female
Male
Female
Male
Male
Female
Male
Female
Female
Female
Female
Female
Female
Female
Female
Unknown
HUS
Bloody diarrhea
Bloody diarrhea
Bloody diarrhea
HUS
HUS
HUS
HUS
HUS
HUS
HUS
HUS
HUS
HUS
HUS
HUS
—
—
German
German
German
German
French
French
French
French
French
French
French
French
French
French
French
3066 | www.pnas.org/cgi/doi/10.1073/pnas.1121491109
Grad et al.
B
A
422387, 1546241
Ec11-5538
Ec11-4632.2
3621338
170476, 2029740
87
67
Ec11-4632.4
Ec11-4632.5
1256852, 2937308, 5226522
93
Ec11-4632.3
91
79
04-8351
09-7901
01-09591
Ec11-4632.1
95
Ec11-4623
C227-11
Isolates from single individual
55989
Ec11-5536
551216, 1262666, 2831655, 2932413
Ec11-4522
4114250, 4243327, 4660485, 4807228
2276417
1568661
Ec11-5537
Ec11-4404
1096014
65
224851
2252380
63
E1167
C236-11
C227-11
German outbreak isolates
Ec11-3677
Ec11-3798
Ec09-7901
c
EC04-8351
Historical isolates
0.05
0.002
Materials and Methods and Figs. S2–S4). We found no other evidence of gene gain or loss.
Discussion
In this study, we perform whole-genome sequencing of multiple
isolates from the 2011 outbreaks of E. coli O104:H4 in France and
Germany to identify differences among isolates that are indistinguishable by standard molecular epidemiological tools. We
find that the isolates are all closely related, and that the German
outbreak isolates have extremely limited diversity, whereas there
is greater diversity among the isolates from the French outbreak.
Several lines of evidence support our finding of extremely
limited diversity among at least a majority of the German outbreak isolates. First, there is minimal diversity among the four
independent isolates reported here (see Table 1 and Materials and
Methods for description of the background of the isolates). Second, a previous analysis of two other isolates identified no SNPs
between them (10). The chance of detecting a subpopulation that
comprises 40% of the overall population using six randomly selected isolates is 95% [1 − (1 − 0.4)6 = 0.95]. Even in the absence
of the two isolates from the independent analysis, the likelihood
of detecting a subpopulation of 40% of the total population with
four isolates is 87% [1 − (1 − 0.4)4 = 0.87]. Thus, our sample size
is sufficient to detect, with high probability, variants present as
a majority or large minority of all isolates. Third, eight isolates
from the German outbreak with sequence in GenBank (GOS1,
GOS2, H112180540, H112180541, H112180280, H112180282,
H112180283, and LB226692) share identical sequence to TY2482
at the sites of each SNP position described in this study. Although
it is impossible to exclude the possibility of unsampled diversity in
the German outbreak, our findings argue that a majority of the
population is extremely closely related.
Using the framework of the trace-back epidemiology that links
the two outbreaks to the 2009 shipment of fenugreek seeds,
several hypotheses can explain the surprising findings that there
is greater diversity of E. coli O104:H4 in the much smaller
Grad et al.
French outbreak than the German outbreak, and that the outbreak isolates from Germany appear to be nested within the
diversity of the French outbreak (Fig. 2).
One hypothesis is that the limited diversity reflects a stochastic
bottleneck in at least the sampled part of the E. coli pathogen
population in Germany compared with France. As we found no
evidence for positive or purifying selection in the SNPs, the
bottleneck we propose represents a random process that purged
most of the diversity. The limited diversity observed within an
individual suggests the hypothesis that the bottleneck in the
German outbreak could represent contamination from a single
infected human at the sprout farm in Germany. Consistent with
this hypothesis, three employees were confirmed as early cases of
E. coli O104:H4 infection, including two asymptomatic shedders,
dating to around the time of the reported start of the outbreak in
early May 2011 (16). In principle, the limited diversity in Germany could also result from partially successful measures to
disinfect seeds or sprouts at the German sprout farm; however, it
appears that no specific disinfection procedures were applied,
apart from routine hygiene and cleaning of the sprout preparation area (16). Analysis of any isolates available from the earliest
stages of the outbreak, including those from infected employees
or sprouts, would allow for direct testing of these hypotheses.
Broader sampling from the outbreak in Germany may help determine the extent to which the outbreak in Germany reflects
contamination from a single individual, and whether there is
evidence for subpopulations with additional diversity.
A second hypothesis is that although substantial diversity was
present in the original bacterial source population, it was unevenly distributed, with a more diverse population, perhaps
reflecting heavier contamination, affecting seeds sent to France
more than those sent to Germany. As a far greater amount of
seeds (10,500 kg) went to the German distributor that supplied
the establishment identified as the source of the German outbreak and only 400 kg went to the English distributor that supplied the 50-g seed packets believed to be the source of the
PNAS | February 21, 2012 | vol. 109 | no. 8 | 3067
MICROBIOLOGY
Fig. 1. (A) Bootstrap consensus maximum-likelihood phylogeny using the 21 SNPs, based on 500 bootstraps and rooted on the 2004 and 2009 isolates. No
branch lengths are provided for the 2004 and 2009 isolates because this phylogeny is generated only from the 21 SNPs from the two outbreaks. The isolates
associated with the German outbreak are C236-11, C227-11, Ec11-3677, and Ec11-3798. The 2004 and 2009 isolates are Ec04-8351 and Ec09-7901, respectively.
The remaining isolates are associated with the French outbreak. Ec11-4632.1 to Ec11-4632.5 represent the five isolates from a single individual. The black
numbers at nodes indicate bootstrap support. The maroon numbers along the branches indicate the locations, with respect to the TY2482 genome, of the
SNPs that define each branch. (B) Bootstrap consensus maximum-likelihood phylogeny using SNPs derived from whole-genome alignment of assemblies of
C227-11, 55989, 01–09591, Ec04-8351, Ec09-7901, and the commensal E. coli E1167, as described in Materials and Methods. The black numbers at nodes
indicate bootstrap support.
Table 2. SNPs identified within E. coli O104:H4 outbreak isolates
SNP position
170476
224851
422387
551216
1096014
1256852
1262666
1546241
1568661
2029740
2252380
2276417
2831655
2932413
2937308
3621338
4114250
4243327
4660485
4807228
5226522
Gene/region
Cyclic diguanylate phosphodiesterase
domain-containing protein
Calcium proton antiporter
Primary amine oxidase
HTH-type transcriptional regulator
Aromatic amino acid transporter
(tyrosine specific)
Wzy
NeuD family sugar
O-acyltransferase
Phosphatase yfbT
dedA
Intergenic between sulfite
reductace hemoprotein β-component
and phosphoadenosine phosphosulfate
reductase
L-asparaginase 2
Type 3 restriction enzyme/helicase
OR PstII subunit
sn-glycerol-3-phosphate
Transport system permease ugpE
Hypothetical protein
di-haem Cytochrome c peroxidase
family protein
Intergenic: between soxR
redox-sensitive transcriptional
activator and yjcD putative permease
3-Isopropylmalate dehydratase
large subunit
Lysine decarboxylase 2
iniconductance mechanosensitive
channel
Intergenic: between hypothetical
protein and citrate synthase
Conserved hypothetical protein
Isolates
SNP*
Substitution
Ec11-4632 C1-C5
G→T
Ser150Ile
C227-11
Ec11-5538
Ec11-4522
C236-11
A→T
C→T
G→A
C→T
Glu366Val
Ser703Leu
Synonymous
Synonymous
Ec11-4623; Ec11-4632 C1-C5; Ec11-5536
Ec11-4522
G→A
A→T
Arg361Gln
Glu132Asp
Ec11-5538
Ec04-8351; Ec09-7901; Ec11-4522;
Ec11-4623; Ec11-4632 C1-C5; Ec11-5536;
Ec11-5538
Ec11-4632 C1-C5
G→T
G→T
Ala48Ser
Gly145Val
C→A
N/A
Ec04-8351; Ec09-7901; Ec11-4404;
Ec11-4522; Ec11-4623; Ec11-4632 C1-C5;
Ec11-5536; Ec11-5537; Ec11-5538
Ec11-4404
T→C
Leu271Pro
A→C
Asp393Ala
Ec11-4522
C→A
Cys99Stop
Ec11-4522
Ec11-4623; Ec11-4632 C1-C5;
Ec11-5536
Ec11-4632 C2-C5
C→A
A→C
Arg73Ser
Synonymous
T→A
N/A
Ec11-5537
T→G
Ile107Ser
Ec11-5537
Ec11-5537
A→C
C→A
Lys367Gln
Synonymous
Ec11-5537
T→C
N/A
Ec11-4623; Ec11-4632 C1-C5; Ec11-5536
A→T
Asn2476Tyr
SNP position is with reference to the TY2482 genome. N/A, not applicable, as SNP not in coding sequence.
*SNP base differences are called with respect to the coding strand, rather than with respect to the Fasta sequence for TY2482.
French outbreak (7), this hypothesis requires the low probability
event that seeds with the higher diversity E. coli population
happened to be in the smaller-sized shipments. Characterization
of E. coli O104:H4 populations found on other seeds from this
shipment may help to assess this hypothesis. To our knowledge,
no such populations have yet been described.
Finally, a third hypothesis is that the difference in diversity
reflects unknown environmental or other constraints that influenced rates of accumulation of diversity once the bacteria arrived
in each country. For example, it is possible that differences in
sprouting conditions between the German sprout farm and the
French community center could have led to differences in diversity. These differences in conditions include use of well-water at a
temperature of 20 °C in the sprout farm in Germany (16), compared with tap water at ambient temperature (between 12 and 28 °C)
in the French outbreak (2). Seeds in France were also germinated
for about 1.5 d longer. Testing rates of accumulation of SNPs
under various conditions may help to assess this possibility.
Using next-generation sequencing methods, we have been able
to reveal variation at a single nucleotide level within genome
sequences from a point-source outbreak, all within a set of isolates
3068 | www.pnas.org/cgi/doi/10.1073/pnas.1121491109
that are identical by classic typing techniques. Highly accurate
sequencing and SNP identification can overcome the noise from
sequencing error and discern phylogenetic signal, which may, as in
this case, depend on a small number of nucleotides. As demonstrated by the multiple independent sequencing efforts related to
this E. coli O104:H4 outbreak (9, 10, 13, 15), and also epidemiological investigations of other infectious diseases (17–19), genomic
epidemiology is likely to become the standard strategy in molecular epidemiology as the cost of sequencing continues to decline
and technology becomes more widely accessible.
The determination of genome sequence is already recognized
as a vital part of investigating any new outbreak, to place the
pathogen in context and gain insight into its origins and the basis
of its pathogenicity. Together with other recent work (17–19),
this study argues strongly for multiple genome sequences to
understand patterns of transmission within an outbreak. Such
analyses can already be conducted in a matter of days, and
technological advances will only improve our ability to perform
them in real time. The advantages of whole genome data include
greater resolution than classic techniques for outbreak investigation, such as pulsed-field gel electrophoresis, and a body
Grad et al.
A
B
Data and inferences
Physical separation hypothesis
Seed population
Seed population
?
Possible unsampled diversity
in German outbreak
Diversity seen in the
French outbreak
C
Diversity seen in the
German outbreak
D
Variable mutation rate hypothesis
Diversity seen in the
German outbreak
Bottleneck hypothesis
Diversity in the seed population
Seed population
Mutation rate µ1
Diversity seen in the
French outbreak
Mutation rate µ2
Bottleneck
Diversity seen in the
French outbreak
Diversity seen in the
German outbreak
Diversity seen in the
French outbreak
Fig. 2. Schematic of hypotheses to explain differences in E. coli SNP diversity seen in the French and German outbreaks. (A) At minimum, the contaminating
population that gave rise to both the French and German outbreaks was polymorphic at location 1568661, and possibly other sites, indicating at least two
types of genotypes in the original contaminating population (represented in green and blue). In the samples presented here, there is greater diversity in the
French outbreak E. coli O104:H4 population than observed in the German outbreak population. Although the probability that our sample represents the
majority of the German outbreak is high (see main text), unsampled diversity may exist in a minority of cases in the German outbreak. (B) By the physical
separation hypothesis, there was uneven distribution of the diversity in the original contaminating E. coli O104:H4 population, with the 50-g seed packets
that led to the French outbreak containing a greater degree of diversity than was present in the 75-kg of seed sent to the German sprout farm (7). (C) By the
variable mutation rate hypothesis, the original seed population, comprising at least two genotypes, may have mutated more quickly along the route to the
French outbreak because of environmental or other factors. (D) By the bottleneck hypothesis, the French outbreak diversity represents the original diversity
present in the contaminated seeds. Either a subset or overlapping set of strains that led to the French outbreak were sent to the German sprout farm,
followed by a bottleneck that restricted diversity in the German outbreak. A bottleneck could have taken place from the time of separation of the seeds from
the original shipment to the German and English seed distributors through germination in the sprout facility. For discussion of factors favoring each hypothesis, see the main text.
of data amenable to analysis with well-developed and understood
phylogenetic methods. As this example demonstrates, the results
of such analysis, combined with traditional epidemiology, can
raise novel epidemiologic hypotheses and questions that are
available only through sequencing of multiple isolates.
Genome Sequencing. We used a multiplatform strategy, generating an average of 146-fold sequence coverage on the Illumina platform, supplemented
with data from 454 and Pacific Bioscience platforms for specific analyses. For
details of the sequencing methods and genome assembly, see SI Materials
and Methods.
Materials and Methods
SNP Prediction and Validation. SNP calling was performed using our analysis
pipeline [GATK v1.0.6011 (21)] based on alignments of paired-end read data
(101 sequences from both ends of 180-bp insert fragments on the Illumina
platform) to the TY2482 strain. Potential SNPs from the Illumina sequences
were called by GATK Unified Genotyper (22), filtering the data according to
the following parameters: >90% agreement among reads; at least five unambiguously mapped reads; no greater than 50% mapping ambiguity;
insertions and deletions were ignored. Over 97% of the bases in the genome
of each outbreak isolate fulfilled these criteria. Bases were identified that
have the highest computational likelihood for calling a base as either
agreement to the reference or a SNP. Only SNPs at locations where equally
high-confidence calls could be made in all outbreak isolates were included in
the analysis. At 54 sites, all outbreak and historical isolates showed the same
sequence as each other but disagreed with the TY2482 reference genome; we
did not identify these sites as SNPs and use them as discriminatory markers
because they may represent errors in the reference sequence as opposed to
true SNPs (Fig. S1 and Table S2). See SI Materials and Methods for details of
454-based genome sequencing and SNP validation and PCR-based validation.
Strains Sequenced in This Study. Isolates include 4 linked to the outbreak
centered in Germany, 11 from the outbreak in the Bordeaux area of France (of
which 5 are from a single individual), and 2 2004 and 2009 Shiga toxin-producing O104:H4 isolates from France. The German outbreak isolates were
linked by travel to Germany and timing of the cases. C227-11 derives from a
68-y-old woman originally from Hamburg, Germany, who was in Denmark when
she fell ill; the isolate was obtained on May 18. Note that a genome sequence for
this isolate was previously reported (15). To ensure consistency in our analyses,
we independently sequenced this isolate and use the genome sequence we
generated for the studies reported here. C236-11 was isolated from a 23-y-old
man from Southern Denmark, which borders Germany, without confirmed
travel to Germany; the isolate was obtained on May 21. Ec11-3677 derives from
a 31-y-old German woman who had spent 2 wk in Northern Germany (May 5–21,
2011) and who was traveling in France at the time of illness on May 21. Ec113798 was isolated from a 55-y-old French man who traveled in Northern Germany between May 8 and 12, 2011, and had returned to France when he
became ill on May 21. The French outbreak isolates (Ec11-4404, Ec11-4522,
Ec11-4623, Ec11-4632_C1-C5, Ec11-5536, Ec11-5537, Ec11-5538) were collected
from individuals in the same community near Bordeaux, all of whom were
known to eat sprouts at a single event on June 8, 2011 (2). Ec04-8351 and Ec097901 were isolated from the stool of infected individuals in France in 2004 and
2009 and represent historical O104:H4 isolates (20) (Table 1).
Grad et al.
Phylogenetic Analysis. To study the phylogenetic relationship among the
outbreak isolates, we created a single sequence for each isolate consisting of
the genotype at the 21 SNP sites and used these data as input sequence to
Mega (23). A maximum-likelihood tree was generated using the Kimura
PNAS | February 21, 2012 | vol. 109 | no. 8 | 3069
MICROBIOLOGY
Diversity seen in the
German outbreak
two-parameter model with 500 bootstraps and rooted on the branch leading to the 2004 and 2009 isolates. To study the relationship between the
outbreak and historical isolates, we first aligned whole-genome assemblies
of C227-11, 55989, 01–09591, 04–8351, 09–7901, and the commensal E. coli
E1167 using progressiveMauve (24). We selected SNPs from this alignment
that contain unambiguous bases for all isolates, are in regions that align,
and have at least 90% agreement in a sliding 100-bp window around each
SNP. These SNPs were used to generate a maximum-likelihood tree using the
Kimura two-parameter model with 500 bootstraps and rooted on E1167.
ACKNOWLEDGMENTS. We thank Flemming Scheutz from the World Health
Organization Collaborating Centre for Reference and Research on Escherichia coli and Klebsiella, Department of Microbiological Surveillance and Research, Statens Serum Institut, Copenhagen, Denmark for the valuable
contribution of strains and discussion; R. T. Chandler for his insights; all
members of the Broad Biological Samples Platform and the Genome Sequencing Platform; Lynne Aftuck, Scott Anderson, Patrick Cahill, Marc Chevr-
ette, Julio Diaz, Danielle Dionne, Sheli Dookran, Rachel Erlich, Stephanie
Grandbois, Lisa Green, Andrew Hollinger, Laurie Holmes, Andrew Hoss,
Edward Kelliher, Sharon Kim, Purnima Kompella, Tony LaCasse, Matthew
Lee, Niall J. Lennon, Dana Robbins, Alyssa Rosenthal, Elizabeth Ryan, Brian
Sogoloff, Todd Sparrow, Alvin Tam, Austin Tzou, Cole Walsh, and Emily
Wheeler for sequencing and technical support; and Lucia Alvarado-Balderrama, Aaron Berlin, Gary Gearin, Sante Gnerre, Giles Hall, Alma Imavovic,
David B. Jaffe, Annie Lui, Iain MacCallum, J. Pendexter Macdonald, Matthew
Pearson, Margaret Priest, Dariusz Przbylski, Andrew Roberts, Filipe J. Ribeiro,
Sakina Saif, Ted Sharpe, Terry Shea, Narmada Shenoy, and Shuangye Yin for
computational and analysis support. This project has been funded in part with
federal funds from the National Institute of Allergy and Infectious Diseases,
National Institutes of Health, Department of Health and Human Services, under Contract HHSN272200900018C (to B.W.B.), and the Infectious Disease Program of the Broad Institute; National Institute of General Medical Sciences
Award U54GM088558 (to M.L. and W.P.H.); National Institutes of Allergy
and Infectious Disease T32 Grant AI007061 (to Y.H.G.); Danish Council for
Strategic Research Grant 09-063070 (to K.A.K.); and the Institut de Veille Sanitaire (F.-X.W. and E.B.).
1. Frank C, et al.; HUS Investigation Team (2011) Epidemic profile of Shiga-toxin-producing Escherichia coli O104:H4 outbreak in Germany. N Engl J Med 365:1771–1780.
2. Gault G, et al. (2011) Outbreak of haemolytic uraemic syndrome and bloody diarrhoea due to Escherichia coli O104:H4, south-west France, June 2011. Euro Surveill
16:pii: 19905.
3. Bielaszewska M, et al. (2011) Characterisation of the Escherichia coli strain associated
with an outbreak of haemolytic uraemic syndrome in Germany, 2011: A microbiological study. Lancet Infect Dis 11:671–676.
4. Frank C, et al.; HUS investigation team (2011) Large and ongoing outbreak of haemolytic uraemic syndrome, Germany, May 2011. Euro Surveill, 16: pii: 19878.
5. Scheutz F, et al. (2011) Characteristics of the enteroaggregative Shiga toxin/verotoxinproducing Escherichia coli O104:H4 strain causing the outbreak of haemolytic uraemic
syndrome in Germany, May to June 2011. Euro Surveill 16:pii: 19889.
6. Jansen A, Kielstein JT (2011) The new face of enterohaemorrhagic Escherichia coli
infections. Euro Surveill 16:pii: 19898.
7. European Food Safety Authority (2011) Tracing seeds, in particular fenugreek (Trigonella foenum-graecum) seeds, in relation to the Shiga toxin-producing E. coli (STEC)
O104:H4 2011 Outbreaks in Germany and France. Available at http://www.efsa.
europa.eu/en/supporting/doc/176e.pdf. Accessed August 4, 2011.
8. Mariani-Kurkdjian P, Bingen E, Gault G, Jourdan-Da Silva N, Weill FX (2011) Escherichia coli O104:H4 south-west France, June 2011. Lancet Infect Dis 11:732–733.
9. Rohde H, et al.; E. coli O104:H4 Genome Analysis Crowd-Sourcing Consortium (2011)
Open-source genomic analysis of Shiga-toxin-producing E. coli O104:H4. N Engl J Med
365:718–724.
10. Brzuszkiewicz E, et al. (2011) Genome sequence analyses of two isolates from the
recent Escherichia coli outbreak in Germany reveal the emergence of a new pathotype: Entero-Aggregative-Haemorrhagic Escherichia coli (EAHEC). Arch Microbiol 193:
883–891.
11. Armstrong GL, Hollingsworth J, Morris JG, Jr. (1996) Emerging foodborne pathogens:
Escherichia coli O157:H7 as a model of entry of a new pathogen into the food supply
of the developed world. Epidemiol Rev 18:29–51.
12. Touchon M, et al. (2009) Organised genome dynamics in the Escherichia coli species
results in highly diverse adaptive paths. PLoS Genet 5:e1000344.
13. Mellmann A, et al. (2011) Prospective genomic characterization of the German enterohemorrhagic Escherichia coli O104:H4 outbreak by rapid next generation sequencing technology. PLoS ONE 6:e22751.
14. Rocha EP, et al. (2006) Comparisons of dN/dS are time dependent for closely related
bacterial genomes. J Theor Biol 239:226–235.
15. Rasko DA, et al. (2011) Origins of the E. coli strain causing an outbreak of hemolyticuremic syndrome in Germany. N Engl J Med 365:709–717.
16. Bundesinstitut für Risikobewertung (2011) Relevance of sprouts and germ buds as
well as seeds for sprout production in the current EHEC O104:H4 outbreak event in
May and June 2011. Updated Opinion No. 23/2011 of BfR, 5 July 2011. Available at
http://www.bfr.bund.de/cm/349/relevance_of_sprouts_and_germ_buds_as_well_as_
seeds_for_sprouts_production_in_the_current_ehec_o104_h4_outbreak_event_in_may_
and_june_2011.pdf. Accessed December 28, 2011.
17. Gardy JL, et al. (2011) Whole-genome sequencing and social-network analysis of
a tuberculosis outbreak. N Engl J Med 364:730–739.
18. Harris SR, et al. (2010) Evolution of MRSA during hospital transmission and intercontinental spread. Science 327:469–474.
19. Rasko DA, et al. (2011) Bacillus anthracis comparative genome analysis in support of
the Amerithrax investigation. Proc Natl Acad Sci USA 108:5027–5032.
20. Monecke S, et al. (2011) Presence of Enterohemorrhagic Escherichia coli ST678/O104:
H4 in France prior to 2011. Appl Environ Microbiol 77:8784–8786.
21. McKenna A, et al. (2010) The Genome Analysis Toolkit: a MapReduce framework for
analyzing next-generation DNA sequencing data. Genome Res 20:1297–1303.
22. DePristo MA, et al. (2011) A framework for variation discovery and genotyping using
next-generation DNA sequencing data. Nat Genet 43:491–498.
23. Tamura K, et al. (2011) MEGA5: Molecular evolutionary genetics analysis using
maximum likelihood, evolutionary distance, and maximum parsimony methods. Mol
Biol Evol 28:2731–2739.
24. Darling AE, Mau B, Perna NT (2010) progressiveMauve: Multiple genome alignment
with gene gain, loss and rearrangement. PLoS ONE 5:e11147.
3070 | www.pnas.org/cgi/doi/10.1073/pnas.1121491109
Grad et al.
Corrections and Retraction
CORRECTIONS
MICROBIOLOGY
Correction for “Genomic epidemiology of the Escherichia coli
O104:H4 outbreaks in Europe, 2011,” by Yonatan H. Grad, Marc
Lipsitch, Michael Feldgarden, Harindra M. Arachchi, Gustavo C.
Cerqueira, Michael FitzGerald, Paul Godfrey, Brian J. Haas,
Cheryl I. Murphy, Carsten Russ, Sean Sykes, Bruce J. Walker,
Jennifer R. Wortman, Sarah Young, Qiandong Zeng, Amr
Abouelleil, James Bochicchio, Sara Chauvin, Timothy DeSmet,
Sharvari Gujja, Caryn McCowan, Anna Montmayeur, Scott
Steelman, Jakob Frimodt-Møller, Andreas M. Petersen, Carsten
Struve, Karen A. Krogfelt, Edouard Bingen, François-Xavier
Weill, Eric S. Lander, Chad Nusbaum, Bruce W. Birren, Deborah
T. Hung, and William P. Hanage, which appeared in issue 8,
February 21, 2012, of Proc Natl Acad Sci USA (109:3065–3070;
first published February 6, 2012; 10.1073/pnas.1121491109).
The authors note that on page 3066, right column, second full
paragraph, lines 6–7, “Of the 14 SNPs within coding regions, 4
(28.6%) are synonymous” should instead appear as “Of the 14
SNPs within coding regions, 5 (36%) are synonymous.”
Additionally, the authors note that Table 2 appeared incorrectly. The corrected table and its legend appear below.
SNP position
170476
224851
422387
551216
1096014
1256852
1262666
1546241
1568661
2029740
2252380
2276417
2831655
2932413
2937308
3621338
4114250
4243327
4660485
4807228
5226522
Gene/region
Cyclic diguanylate phosphodiesterase
domain-containing protein
Calcium proton antiporter
Primary amine oxidase
HTH-type transcriptional regulator
Aromatic amino acid transporter
(tyrosine specific)
Wzy
NeuD family sugar O-acyltransferase
Phosphatase yfbT
dedA
Intergenic: between sulfite reductace hemoprotein
β-component and phosphoadenosine
phosphosulfate reductase
L-asparaginase 2
Type 3 restriction enzyme/helicase OR PstII subunit
sn-gluycerol-3-phosphate transport system
permease ugpE
Hypothetical protein
di-haem cytochrome c peroxidase
family protein
Intergenic: between soxR redox-sensitive
transcriptional activator and yjcD
putative permease
3-isopropylmalate dehydratase large subunit
Lysine decarboxylase 2
Miniconductance mechanosensitive channel
Intergenic: between hypothetical protein and
citrate synthase
Conserved hypothetical protein
Isolates
SNP*
Substitution
Ec11-4632 C1-C5
G→T
Ser150Ile
C227-11
Ec11-5538
Ec11-4522
C236-11
A→T
C→T
G→A
C→T
Glu85Val
Ser703Leu
Synonymous
Synonymous
Ec11-4623; Ec11-4632 C1-C5; Ec11-5536
Ec11-4522
Ec11-5538
Ec04-8351; Ec09-7901; Ec11-4522;
Ec11-4623; Ec11-4632 C1-C5;
Ec11-5536; Ec11-5538
Ec11-4632 C1-C5
G→A
A→T
G→T
G→T
Arg361Gln
Glu132Asp
Ala48Ser
Gly145Val
C→A
N/A
Ec04-8351; Ec09-7901; Ec11-4404; Ec11-4522;
Ec11-4623; Ec11-4632 C1-C5; Ec11-5536;
Ec11-5537; Ec11-5538
Ec11-4404
Ec11-4522
T→C
Synonymous
A→C
C→A
Asp393Ala
Cys189Stop
Ec11-4522
Ec11-4623; Ec11-4632 C1-C5; Ec11-5536
C→A
A→C
Arg73Ser
Synonymous
Ec11-4632 C2-C5
T→A
N/A
Ec11-5537
Ec11-5537
Ec11-5537
Ec11-5537
T→G
A→C
C→A
T→C
Ile107Ser
Lys367Gln
Synonymous
N/A
Ec11-4623; Ec11-4632 C1-C5; Ec11-5536
A→T
Asn2476Tyr
SNPs identified within E. coli O104:H4 outbreak isolates. SNP position is with reference to the TY2482 genome. N/A, not applicable, as SNP not in coding
sequence.
*SNP base differences are called with respect to the coding strand, rather than with respect to the FASTA sequence for TY2482.
www.pnas.org/cgi/doi/10.1073/pnas.1203955109
www.pnas.org
PNAS | April 3, 2012 | vol. 109 | no. 14 | 5547–5548
CORRECTIONS AND
RETRACTION
Table 2. SNPs identified within E. coli O104:H4 outbreak isolates
DEVELOPMENTAL BIOLOGY
Correction for “Solute carrier family 3 member 2 (Slc3a2) controls yolk syncytial layer (YSL) formation by regulating microtubule networks in the zebrafish embryo,” by Aya Takesono,
Julian Moger, Sumera Faroq, Emma Cartwright, Igor B. Dawid,
Stephen W. Wilson, and Tetsuhiro Kudoh, which appeared
in issue 9, February 28, 2012, of Proc Natl Acad Sci USA
(109:3371–3376; first published February 13, 2012; 10.1073/
pnas.1200642109).
The authors note that the author name Sumera Faroq should
instead appear as Sumera Farooq. The corrected author line
appears below. The online version has been corrected.
Aya Takesono, Julian Moger, Sumera Farooq,
Emma Cartwright, Igor B. Dawid, Stephen W. Wilson,
and Tetsuhiro Kudoh
www.pnas.org/cgi/doi/10.1073/pnas.1203335109
MEDICAL SCIENCES
Correction for “Wounding enhances epidermal tumorigenesis by
recruiting hair follicle keratinocytes,” by Maria Kasper, Viljar
Jaks, Alexandra Are, Åsa Bergström, Anja Schwäger, Nick
Barker, and Rune Toftgård, which appeared in issue 10, March
8, 2011, of Proc Natl Acad Sci USA (108:4099–4104; first published February 14, 2011; 10.1073/pnas.1014489108).
The authors note that Jessica Svärd and Stephan Teglund
should be added to the author list between Anja Schwäger and
Nick Barker. Jessica Svärd should be credited with contributing
new reagents/analytical tools. Stephan Teglund should be credited with contributing new reagents/analytical tools. The corrected
author and affiliation lines, and author contributions appear below. The online version has been corrected.
Maria Kaspera,1, Viljar Jaksa,b,1, Alexandra Area, Åsa
Bergströma, Anja Schwägera, Jessica Svärda, Stephan
Teglunda, Nick Barkerc,2, and Rune Toftgårda,3
a
Center for Biosciences and Department of Biosciences and Nutrition,
Karolinska Institutet, Novum, 141 83 Huddinge, Sweden; bInstitute of
Molecular and Cell Biology and Estonian Biocentre, University of Tartu,
51010 Tartu, Estonia; and cHubrecht Institute, Koninklijke Nederlandse
Akademie van Wetenschappen and University Medical Center Utrecht,
3584CT Utrecht, The Netherlands
Author contributions: M.K., V.J., and R.T. designed research; M.K., V.J., A.A., Å.B., and A.S.
performed research; J.S., S.T., and N.B. contributed new reagents/analytic tools; M.K., V.J.,
A.A., Å.B., A.S., and R.T. analyzed data; and M.K., V.J., and R.T. wrote the paper.
www.pnas.org/cgi/doi/10.1073/pnas.1203264109
RETRACTION
IMMUNOLOGY
Retraction for “The antibody zalutumumab inhibits epidermal
growth factor receptor signaling by limiting intra- and intermolecular flexibility,” by Jeroen J. Lammerts van Bueren, Wim K.
Bleeker, Annika Brännström, Anne von Euler, Magnus Jansson,
Matthias Peipp, Tanja Schneider-Merck, Thomas Valerius, Jan
G. J. van de Winkel, and Paul W. H. I. Parren, which appeared in
issue 16, April 22, 2008, of Proc Natl Acad Sci USA (105:6109–
6114; first published April 21, 2008; 10.1073/pnas.0709477105).
The undersigned authors wish to note the following: “In our
study we employed Protein Tomography (PT), an electron tomography method commercialized by Sidec AB, for structural
analysis of proteins. Following our publication, doubts were
raised with respect to the validity of Sidec PT, and we therefore
conducted an extensive validation study with Sidec AB to determine the fidelity of PT for structural analysis of therapeutic
antibodies and antibody–antigen complexes in solution. In two
independent double blind experiments, PT was found to be highly
unreliable in distinguishing structural features of the molecules
and complexes studied. First, approximately 90% of the identified
protein density maps could not be interpreted due to complex
morphology or low quality. Second, among the remaining objects,
a high number of the protein images observed did not match
with sample composition resulting in misinterpretation of sample identities. These disappointing results led us to reanalyze our
previously acquired PT data in situ using a weighted backprojection reconstruction method with IMOD (1) rather than
COMET (2), which also indicated our previous analysis to be
unreliable. With the current PT methods, we were thus not able
to validate tomograms neither obtained from vitrified proteins
in solution nor aldehyde-fixed, stained cells. Therefore the undersigned authors no longer feel confident of the PT data presented in Figs. 3, 4, 5, and S4 of our study. The authors stand
behind the supporting biochemical data provided in the manuscript and believe the proposed model to be plausible, the validity
of which, however, should be addressed with other methods. We
found it important to notify our colleagues of the specific technical flaws in our publication and apologize for any inconvenience
caused. We hereby retract this manuscript.”
Jeroen J. Lammerts van Bueren
Wim K. Bleeker
Annika Brännström
Magnus Jansson
Matthias Peipp
Tanja Schneider-Merck
Thomas Valerius
Jan G. J. van de Winkel
Paul W. H. I. Parren
1. Kremer JR, Mastronarde DN, McIntosh JR (1996) Computer visualization of threedimensional image data using IMOD. J Struct Biol 116:71–76.
2. Skoglund U, Ofverstedt LG, Burnett RM, Bricogne G (1996) Maximum-entropy threedimensional reconstruction with deconvolution of the contrast transfer function: a test
application with adenovirus. J Struct Biol 117:173–188.
www.pnas.org/cgi/doi/10.1073/pnas.1203736109
5548 | www.pnas.org